ResearchBenchmarksImage Classification on CIFAR-100 (test) (Symmetric/Asymmetric Noise Corruption)For AI assistantsMarkdownContent indexImage Classification on CIFAR-100 (test) (Symmetric/Asymmetric Noise Corruption)Follow79.3Accuracy (Symmetric 20% Noise)NGC75.86876.75977.6578.541Mar 8, 2022Evaluation resultsMethodMethodLinksAccuracy (Symmetric 20% Noise)Accuracy (Symmetric 90% Noise)Accuracy (Asymmetric 40% Noise)NGCTraining Pipeline=One-...Training Pipeline=One-stage2022.0379.329.8—ProtoMixTraining Pipeline=One-...Training Pipeline=One-stage2022.0379.129.348.8Sel-CL+Training Pipeline=One-...Training Pipeline=One-stage, Fine-tuning Augmentation=Standard2022.0376.435.574.2Sel-CL+Training Pipeline=One-...Training Pipeline=One-stage, Fine-tuning Augmentation=AugMix2022.037635.474.2